Guardian Ai

• 

  Landing page

• 

• 

• 

  Role Selection

Inspiration

With the alarming rise in online scams, misinformation, emotional burnout, and the growing disconnection among family members in today's tech-saturated world, we were driven to create Guardian AI — a proactive, ethical AI co-pilot for families. Our inspiration came from real-world news: elderly people falling victim to phishing calls, children struggling silently with anxiety, and parents overwhelmed by digital chaos. We knew AI could do more than entertain — it could protect, guide, and uplift entire households.

What it does

Guardian AI is an AI-powered web platform built to support families and communities across generations. It offers: 🔐 Scam Protection Assistant — detects and explains phishing messages using voice or text input. 🧘 Mental Health Voice Journaling — analyzes user tone and responds with support or alerts. 🗣️ Conversational AI Coach — role-specific AI avatars guide users (child, teen, elder, parent) through life advice, education, and emotional wellbeing. 🛡️ Family Safety Dashboard — provides mood trends, scam history, and learning progress per user. 🧩 Accessibility & Theming Controls — with speech navigation, color modes, language switch, and minimize/maximize toggles for inclusivity.

How we built it

Built using React (frontend) and Node.js (backend) on Replit, with Firebase as the authentication and database layer. Integrated OpenAI for intent detection and emotion analysis. Created dedicated feature pages (Scam Protection, Health Reminders, Role Dashboards) for user clarity. Ensured complete responsiveness and accessibility using Tailwind CSS and voice control libraries.

Challenges we ran into

🔐 Integrating secure Firebase Auth for multi-role login (Google/Microsoft) with minimal delay. 🧠 Tuning AI prompts to respond empathetically without being generic or robotic. 🎥 Syncing video avatar replies from Tavus to match user-specific queries and context. 🗺️ Building a flexible routing system that could scale with multiple roles and features. 🌍 Ensuring accessibility features didn’t interfere with the voice-first UX on slower devices.

Accomplishments that we're proud of

Successfully built a multi-role, voice-controlled app that responds like a real human assistant. Designed a scam detection AI that works via voice and helps elderly users understand threats in simple terms. Built the app to run entirely on free-tier services — making it accessible to people in under-resourced areas. Created a project that bridges generations and offers empathy + security through AI.

What we learned

AI isn’t just about speed or smarts — contextual empathy is everything. Building inclusive tech means thinking beyond accessibility — it means understanding emotion, cultural gaps, and psychological safety. Real-time AI video tools (like Tavus) are not just flashy — they build trust in communication. Hackathon deadlines push you to focus on impact-first design, not feature overload.

What's next for Guardian Ai

📱 Mobile-first design: Package the project into a PWA and Android/iOS app with native support. 🧠 Offline Mode: Add caching and local ML to support users with poor internet. 🔎 Deeper Role Customization: Let users customize avatars, voice tone, and content preferences. 📈 Guardian+ Plan: Introduce optional premium features for crisis alerting, multilingual coaching, and therapy bot integration. 🤝 NGO & School Integration: Partner with family services, senior care NGOs, and schools to offer Guardian AI for real-life impact.

Built With

• bolt.new
• firebase-authentication
• firestore
• github
• netlify
• react
• tailwind-css
• typescript
• vite